A little python and matplotlib script I wrote to plot the data from my test devices for any period available in the database.
First thing to note is the massively varying plot on each graph is the external sensor box I also designed using the awesome software engineers cad program “OpenSCAD”, the enclosure design exceeded my expectations, still water safe after almost a year bolted to my shed. I went for a completely exposed base which I figured would get better airflow and allow draining of any liquid and an inner “keep” wall as well as the external vented wall.
Anyway you can see the four internal sensors are all pretty much in agreement, though aqkit went through a period in the kitchen that had better air mainly due to the airflow in the kitchen, and so that actually tracks the external sensor at certain periods, but then also worse air at times whilst cooking.
All sensors in agreement over barometric pressure.
As for the actual temperature reading, the external box is a thin pla 3d printed box mounted on a metal shed, so, kind of a temperature amplifier I would say! The Stevenson screen is key to the thing eh! same wood, same dimensions, same residual humidity in the wood, sealed with N layers of a specific paint… same sensor position in the box, same altitude from the ground, same ground surface (grass, not a Heathrow runway LOL) etc. etc. A very important component! The Enclosure. As for internal temperature think I have not accounted fully for digital self heating yet, need a reliable accurate mercury min/max thermometer and humidity meter for calibration.
You can also see why any air quality algorithm has problems pitching “OK” air when it varies so much in an external setting like this urban yet rural area. The problem for internal sensors is opening all the windows for a day makes all the other readings appear so bad on a linear scale.
Anyway if anyone wants to buy one you can add your name to a list, but the raspberry pi is currently as rare as rocking-horse shit so… a problem…